Sharp vertex connectivity of the Markoff graphs modulo $p$

Jie Ma; Mengxi Yang; Zichen Yang

REVIEW 1 major objections 4 minor 11 references

Sharp vertex connectivity of the Markoff graphs modulo $p$

T0 review · 1 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Every edge of the Markoff graph modulo a prime lies on a cycle.

desk verdict Nice bridge-free result, but a one-character typo in V3 currently breaks the keystone identity. read the letter →

arxiv 2608.07880 v1 pith:C42RHJDZ submitted 2026-08-08 math.NT math.CO

Jie Ma , Mengxi Yang , Zichen Yang This is my paper · ORCID

classification math.NTmath.CO MSC 05C2505C4011D25

keywords MarkoffgraphVietainvolution2-connectivitybridgequotientcuspfinitefielddivergencemethod

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper proves a sharp connectivity statement for the Markoff graphs $G_p$: the graphs whose vertices are nonzero solutions of $x_1^2+x_2^2+x_3^2=x_1x_2x_3$ over the finite field $\mathbb{F}_p$, with edges joining solutions that differ by one of three Vieta involutions. The authors show that every edge of $G_p$ lies on a cycle (Theorem 1.1), equivalently that $G_p$ has no bridges. It follows that whenever $G_p$ is connected, it is 2-vertex-connected: deleting any one vertex leaves the graph connected. Since earlier results show $G_p$ is connected for all sufficiently large primes, this makes $G_p$ 2-connected for all sufficiently large primes. The result is sharp: for every prime $p\ge 7$ the graph has a vertex with exactly two distinct neighbors, so $G_p$ is never 3-connected.

What carries the argument

The central object is the quotient graph $\overline{G}_p = X_p/K$, where $X_p$ is the set of nonzero solutions in $\mathbb{F}_p^3$ and $K=\{1,\sigma_1,\sigma_2,\sigma_3\}$ is the Klein four-group generated by double sign changes such as $(x_1,-x_2,-x_3)$. Cusps are quotient vertices whose representatives have a zero coordinate. The mechanism that carries the argument is a vector-valued divergence: for each regular vertex $x$ and each $i\in\{1,2,3\}$, the paper defines $F_i(x)=(U_i(x),V_i(x))$ from explicit rational functions, with four properties: $F_i(m_i(x))=-F_i(x)$ along a Vieta edge, $F_i$ is invariant under $K$, the three vectors sum to $(0,0)$ at every regular vertex, and $F_i(x)\neq(0,0)$ whenever the Vieta move is nontrivial. Summing the divergences over one side of a bridge forces the bridge's assigned value $D(e)$ to vanish, contradicting the nonzero property; hence cusps must lie on both sides of any quotient bridge. The cusp-lifting property then converts a quotient bridge into a long cycle in $G_p$, proving the main theorem.

What would settle it

Evaluate the identity $F_1(x)+F_2(x)+F_3(x)=(0,0)$ at a regular solution to $x_1^2+x_2^2+x_3^2=x_1x_2x_3$ over a small prime such as $p=5$ or $p=7$; any regular vertex where it fails would disprove Lemma 2.4(3). Alternatively, run an exhaustive search for a bridge in $G_p$ for all primes below $10^6$: Theorem 1.1 predicts that no edge whose removal disconnects the graph exists, so a single bridge found would falsify the main theorem.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is a bridge-free theorem: for every prime $p\ge 5$ and every edge $e$ of the Markoff graph $G_p$, the edge $e$ lies on a cycle. The proof passes to the quotient graph $\overline{G}_p$ obtained by identifying vertices that differ by an even number of coordinate sign changes, a free action of the Klein four-group $(\mathbb{Z}/2\mathbb{Z})^2$. In that quotient, the paper defines cusps — vertices with a zero coordinate — and proves that every connected quotient subgraph containing a cusp lifts to a connected subgraph of $G_p$. The key technical lemma shows that removing a bridge in the quotient leaves cusps on both sides; this is proved by assigning to each oriented edge a two-dimensional vector $F_i$ built from explicit rational functions, showing the divergence at each regular vertex is zero via the identity $F_1+F_2+F_3=(0,0)$, and then observing that a bridge would force a nonzero value $F_i(v)$ to equal $(0,0)$. Lifting the resulting quotient cycle back yields a genuine cycle in $G_p$ containing any prescribed edge.

Load-bearing premise

The proof rests on the algebraic identity $F_1(x)+F_2(x)+F_3(x)=(0,0)$ holding at every regular vertex, which the paper verifies by a direct calculation; if that identity failed for even one regular vertex, the divergence sum over a bridge component would not force the bridge's value to zero, and the bridge-free conclusion would collapse.

Editorial extensions

If this is right

Theorem 1.1: for every prime $p\ge 5$, every edge of $G_p$ lies on a cycle, so $G_p$ has no bridges.
Corollary 1.2: if $G_p$ is connected, then it is 2-vertex-connected, using only the fact that every vertex has degree at most three.
Since $G_p$ is connected for all sufficiently large primes by previous results, it is 2-connected for all sufficiently large primes.
Corollary 1.3: $G_p$ is 2-connected for every prime in $[5,10^6)\cup(3.449\cdot 10^{392},\infty)$.
Remark 1.4: for every prime $p\ge 7$, $G_p$ is not 3-connected, so the 2-connectivity bound is the strongest possible general statement.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The divergence construction looks transferable: any graph built from Vieta involutions on solutions of a generalized Markoff–Hurwitz equation over a finite field, admitting a vector function with the four Lemma 2.4 properties, would be bridge-free by the same summation argument.
Because $G_p$ is never 3-connected for $p\ge 7$, the expander question cannot be settled by exploiting high vertex connectivity; a positive answer would require a different mechanism.
A direct numerical check is available: search for a bridge in $G_p$ for every prime below $10^6$; Theorem 1.1 demands that none exist, giving an independent test of the proof's conclusion.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Here’s my take on arXiv:2608.07880. The headline: the paper proves a nice, natural result—if G_p is connected then it is 2-connected, and the proof works through a clever bridge-free argument. The quotient-by-sign-changes setup and the vector-valued divergence function are genuinely new, and the sharpness observation (not 3-connected for p>=7) is a clean add-on. Credit where due: the paper is honest, self-contained modulo standard facts, and the AI-use declaration is transparent.

That said, the paper as written has a concrete defect. Lemma 2.4(3) states F1+F2+F3=(0,0) on regular vertices. Using the definitions exactly as typeset, this is false. At the regular solution (1,3,4) mod 7, V1=0, V2=5, V3=6, so the sum is 4, not 0. The problem is the denominator of V3: it is printed as x1*x3^2, but the identity forces x1*x2^3 (or something equivalent). With that correction, V3 becomes 2 mod 7 and the sum vanishes at this point. This is the keystone of the divergence argument: Lemma 2.3 sums divergences over one side of a bridge and needs this identity to conclude D(e)=0, then reaches a contradiction via Lemma 2.4(4). So Theorem 1.1 as currently typeset is unsupported.

The good news is that the fix looks like a one-character correction, and the rest of the proof structure is coherent. I did not find other gaps of comparable size. The proof of Lemma 2.4(1) is a bit terse ('by symmetry'), and the denominator in the displayed formula for V1+V2+V3 is garbled, but neither is fatal once the correction is made. The citation pattern is fine; the result does not rest on the authors' own earlier work. The sharpness remark via Cerbu–Gunther–Magee–Peilen is appropriate.

Bottom line: this is a paper worth refereeing, not a desk reject. But it must not be accepted in its current form. A referee should require the authors to fix the V3 denominator, re-verify Lemma 2.4(3) (ideally with a short expansion), and check that Lemma 2.3 still goes through. If they do, the result stands as a solid, sharp contribution to the Markoff graph program.

Referee Report

1 major / 4 minor

Summary. The paper studies the vertex connectivity of the Markoff graphs G_p modulo a prime p. Its main claim (Theorem 1.1) is that every edge of G_p lies in a cycle for every prime p≥5; from this and the sub-cubic degree bound it derives (Corollary 1.2) that a connected G_p is 2-connected, and hence that G_p is 2-connected for all sufficiently large primes. The proof passes to the quotient by the Klein four-group of double sign changes, isolates a class of cusp orbits, and uses a vector-valued function F_i in a divergence argument to show that a bridge in the quotient would force cusps on both components of the bridge. A lifting lemma then converts this quotient statement into the desired cycle in G_p. The paper also records sharpness, noting that G_p is not 3-connected for p≥7.

Significance. If the proof is corrected, this is a natural and valuable strengthening of the known connectivity results for Markoff graphs, directly motivated by the Bourgain–Gamburd–Sarnak expander question. The quotient-by-sign-changes and divergence framework is elegant, and the paper is mostly self-contained. The proof is direct and has no fitted parameters or reliance on the authors' previous results; the explicit range in Corollary 1.3 is a useful bonus. The main caveat is that a displayed identity in the keystone lemma is false as typeset, so the argument as written is not yet sound.

major comments (1)

[Section 2, Lemma 2.4(3)] The identity F1(x)+F2(x)+F3(x)=(0,0) is false as typeset because V3 is defined as V3(x)=(2x3−x1x2)/(x1 x3^2). For the regular solution (1,3,4)∈X_7, one has V1+V2+V3 = 0+5+6 = 4 ≠ 0 in F_7, so the displayed identity fails. Since Lemma 2.3 uses exactly this identity to conclude that the divergence at every vertex of A is (0,0), the proof of Theorem 1.1 as written is unsupported at this point. The displayed common denominator for the V-sum in the proof of (3) is also inconsistent with the definitions of V1, V2, V3. This appears to be a typographical error: changing the denominator of V3 to x1 x2^3 makes the identity true (the same example then gives V1+V2+V3 = 0+5+2 = 0 in F_7), and the subsequent divergence argument appears to go through. Because Lemma 2.4(3) is load-bearing, the manuscript must be corrected and the algebra in the proof of (3) should be written consistently.

minor comments (4)

[Section 2, definitions of U3 and V3] The definition of V3 has an asymmetric denominator compared with U3; presumably V3 should be (2x3−x1x2)/(x1 x2^3) rather than (2x3−x1x2)/(x1 x3^2). Please correct the displayed definition and the subsequent V-sum formula consistently.
[Section 2, proof of Lemma 2.4(3)] The proof says 'A direct calculation gives' and then displays formulas for U1+U2+U3 and V1+V2+V3. The denominator in the second displayed formula is garbled: with the corrected V3 it should be x1 x2^3 x3^3. A short derivation or a comment that the identity is checked by clearing denominators would be helpful.
[Corollary 1.2] The sub-cubic degree bound is stated without proof. It is immediate from the definition that each vertex has at most three distinct neighbors among the three Vieta involutions, with a loop counted once, but a one-sentence justification would make the corollary self-contained.
[Theorem 1.1, opening of proof] The proof of Theorem 1.1 explicitly treats only non-loop edges of G_p. The statement is still complete because a loop is itself a cycle of length one, but this convention could be stated for clarity.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the proof is self-contained and the external inputs are independent standard results.

full rationale

The paper proves Theorem 1.1 directly from explicit graph-theoretic and algebraic lemmas. The key ingredient, Lemma 2.3, is established by constructing vector-valued functions F_i and checking their properties (Lemma 2.4) by direct calculation from the Markoff equation; these properties are stated and verified inside the paper rather than imported from prior work. The lifting and cusp arguments in Lemmas 2.1 and 2.2 are argued from the definitions of the quotient graph and the double sign-change action. The only external inputs are standard, independently published results: Carlitz's point count, Bourgain–Gamburd–Sarnak's connectivity results, Chen's divisibility theorem, and the explicit-range connectivity results of Eddy–Fuchs–Litman–Martin–Tripeny and Brown. None of these citations is authored by the present authors and none is used to define the theorem's conclusion into existence. There are no fitted parameters, no prediction-from-fit steps, and no uniqueness theorem imported from the authors' own earlier work. The skeptical observation that Lemma 2.4(3) may contain a typographical denominator error is a potential correctness issue in the algebra, not a circularity: even if that display needed correction, the argument would still derive the conclusion from an explicit identity rather than from the conclusion itself. Therefore the derivation chain is not circular.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The paper introduces no free parameters or fitted constants. The proof relies on cited external theorems for point counts and connectedness. The new auxiliary functions F_i and the quotient graph are explicit constructions within the proof, not postulated entities with independent empirical handles.

assumptions (4)

domain assumption Carlitz's point count |X_p| = p^2+3p if p≡1 mod4, and p^2-3p if p≡3 mod4
Used in Corollary 1.2 to guarantee |X_p|≥3; cited from [5].
domain assumption G_p is connected for all sufficiently large primes p
Follows from Bourgain-Gamburd-Sarnak giant component plus Chen's divisibility theorem, cited in the introduction; used in the abstract's consequence and Corollary 1.3.
domain assumption G_p is connected for p in [5,10^6) and p>3.449e392
Cited from Brown [4] and Eddy-Fuchs-Litman-Martin-Tripeny [9]; used in Corollary 1.3.
domain assumption Cerbu-Gunther-Magee-Peilen's loop count for G_p
Cited in Remark 1.4 to show G_p is not 3-connected for p≥7.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Sharp vertex connectivity of the Markoff graphs modulo $p$." pith.science (2026). https://pith.science/paper/C42RHJDZ

@misc{pith2026260807880,
  author       = {Pith},
  title        = {Pith review of: Sharp vertex connectivity of the Markoff graphs modulo $p$},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/C42RHJDZ}},
  note         = {Machine review of arXiv:2608.07880}
}

abstract

The Markoff graph $G_p$ modulo a prime $p$ is an undirected graph whose vertices are the nonzero solutions over the finite field $\mathbb{F}_p$ of the normalized Markoff equation \[ x_1^2+x_2^2+x_3^2=x_1x_2x_3, \] where two vertices are adjacent if they differ by a Vieta involution. A major breakthrough of Bourgain, Gamburd, and Sarnak established that $G_p$ contains a giant connected component. Combined with Chen's remarkable divisibility theorem, this implies that $G_p$ is connected for all sufficiently large primes $p$. In the same paper, Bourgain, Gamburd, and Sarnak further asked whether the family $\{G_p\colon p\geq5\}$ forms an expander family. This motivates us to investigate the robustness of connectivity in the Markoff graphs. In this short note, we show that if the Markoff graph $G_p$ is connected, then it is in fact $2$-connected. Consequently, the Markoff graph $G_p$ is $2$-connected for all sufficiently large primes $p$. This is sharp in the sense that $G_p$ is not $3$-connected for any prime $p\geq 7$.

Figures

Figures reproduced from arXiv: 2608.07880 by the authors.

**Figure 1.** A 4-cycle in Gp formed by four vertices lying over a cusp Kx ∈ Gp with x = (0, x2, x3). Set K := {1, σ1, σ2, σ3}. It is easy to verify the relations σ 2 i = 1 for i ∈ {1, 2, 3} and σiσj = σjσi = σk for distinct i, j, k ∈ {1, 2, 3}. Thus we have a group isomorphism K ∼= (Z/2Z) × (Z/2Z). Furthermore, the group K commutes with the three Vieta involutions m1, m2, m3, and acts on Xp freely. Indeed, if a non-identity elem… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

11 extracted references · 11 canonical work pages

[1]

Baragar.The Markoff Equation and Equations of Hurwitz

Arthur B. Baragar.The Markoff Equation and Equations of Hurwitz. PhD thesis, Brown University, 1991

work page 1991
[2]

Markoff triples and strong approximation.Comptes Rendus Math´ ematique, 354(2):131–135, 2016

Jean Bourgain, Alexander Gamburd, and Peter Sarnak. Markoff triples and strong approximation.Comptes Rendus Math´ ematique, 354(2):131–135, 2016

work page 2016
[3]

Strong approximation and Diophantine properties of Markoff triples.Journal of the American Mathematical Society, 39(1):177–204, 2026

Jean Bourgain, Alexander Gamburd, and Peter Sarnak. Strong approximation and Diophantine properties of Markoff triples.Journal of the American Mathematical Society, 39(1):177–204, 2026

work page 2026
[4]

An almost linear time algorithm testing whether the Markoff graph modulopis connected.Research in Number Theory, 11(1), 2025

Colby Austin Brown. An almost linear time algorithm testing whether the Markoff graph modulopis connected.Research in Number Theory, 11(1), 2025. Article 6

work page 2025
[5]

The number of points on certain cubic surfaces over a finite field.Bollettino dell’Unione Matematica Italiana, 12(1):19–21, 1957

Leonard Carlitz. The number of points on certain cubic surfaces over a finite field.Bollettino dell’Unione Matematica Italiana, 12(1):19–21, 1957

work page 1957
[6]

The cycle structure of a Markoff automorphism over finite fields.Journal of Number Theory, 211:1–27, 2020

Alois Cerbu, Elijah Gunther, Michael Magee, and Luke Peilen. The cycle structure of a Markoff automorphism over finite fields.Journal of Number Theory, 211:1–27, 2020

work page 2020
[7]

William Y. Chen. Nonabelian level structures, Nielsen equivalence, and Markoff triples.Annals of Mathematics, 199(1):301– 443, 2024

work page 2024
[8]

Non-planarity of Markoff graphs mod p.Commentarii Mathematici Helvetici, 99(1):111-148, 2024

Matthew de Courcy-Ireland. Non-planarity of Markoff graphs mod p.Commentarii Mathematici Helvetici, 99(1):111-148, 2024

work page 2024

Show all 11 references

[9]

Martin, and Nico Tripeny

Jillian Eddy, Elena Fuchs, Matthew Litman, Daniel E. Martin, and Nico Tripeny. Connectivity of Markoff mod-pgraphs and maximal divisors.Proceedings of the London Mathematical Society, 130(2):e70027, 2025

2025
[10]

Andrey A. Markoff. Sur les formes quadratiques binaires ind´ efinies.Mathematische Annalen, 15:381–406, 1879
[11]

Daniel E. Martin. A new proof of Chen’s theorem for Markoff graphs.Inventiones Mathematicae, 241:623–626, 2025. Email address:jiema@ustc.edu.cn School of Mathematical Sciences, University of Science and Technology of China, Hefei, 230026, People’s Republic of China Yau Mathema...

2025

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

Baragar.The Markoff Equation and Equations of Hurwitz

Arthur B. Baragar.The Markoff Equation and Equations of Hurwitz. PhD thesis, Brown University, 1991

work page 1991

[2] [2]

Markoff triples and strong approximation.Comptes Rendus Math´ ematique, 354(2):131–135, 2016

Jean Bourgain, Alexander Gamburd, and Peter Sarnak. Markoff triples and strong approximation.Comptes Rendus Math´ ematique, 354(2):131–135, 2016

work page 2016

[3] [3]

Strong approximation and Diophantine properties of Markoff triples.Journal of the American Mathematical Society, 39(1):177–204, 2026

Jean Bourgain, Alexander Gamburd, and Peter Sarnak. Strong approximation and Diophantine properties of Markoff triples.Journal of the American Mathematical Society, 39(1):177–204, 2026

work page 2026

[4] [4]

An almost linear time algorithm testing whether the Markoff graph modulopis connected.Research in Number Theory, 11(1), 2025

Colby Austin Brown. An almost linear time algorithm testing whether the Markoff graph modulopis connected.Research in Number Theory, 11(1), 2025. Article 6

work page 2025

[5] [5]

The number of points on certain cubic surfaces over a finite field.Bollettino dell’Unione Matematica Italiana, 12(1):19–21, 1957

Leonard Carlitz. The number of points on certain cubic surfaces over a finite field.Bollettino dell’Unione Matematica Italiana, 12(1):19–21, 1957

work page 1957

[6] [6]

The cycle structure of a Markoff automorphism over finite fields.Journal of Number Theory, 211:1–27, 2020

Alois Cerbu, Elijah Gunther, Michael Magee, and Luke Peilen. The cycle structure of a Markoff automorphism over finite fields.Journal of Number Theory, 211:1–27, 2020

work page 2020

[7] [7]

William Y. Chen. Nonabelian level structures, Nielsen equivalence, and Markoff triples.Annals of Mathematics, 199(1):301– 443, 2024

work page 2024

[8] [8]

Non-planarity of Markoff graphs mod p.Commentarii Mathematici Helvetici, 99(1):111-148, 2024

Matthew de Courcy-Ireland. Non-planarity of Markoff graphs mod p.Commentarii Mathematici Helvetici, 99(1):111-148, 2024

work page 2024